36 results about "Bayesian neural networks" patented technology

This invention relates to the field of aero-engine maintenance engineering and quality assessment technology, and discloses an aero-engine overhaul assessment method based on a physical information Bayesian neural network. The method includes collecting multi-source heterogeneous data from maintenance and testing sites and converting it into key process indicator scores; extracting rotor dynamics, imbalance transmission, and empirical formula features to construct physical constraint regularization terms; establishing an initial Bayesian neural network containing deterministic and variational Bayesian layers; inputting the key process indicator scores into the network, and optimizing parameters by combining the physical constraint regularization terms with adaptive annealing and early stop mechanisms; performing multiple Monte Carlo samplings on the trained network to calculate the mean distribution of the output results and obtain distribution statistical characteristics; calculating the quality score, confidence interval, and attribution warning information based on the distribution statistical characteristics, and outputting an overhaul assessment report. This invention solves the problem that existing pure data models violate physical laws and cannot quantify confidence levels.

The application discloses a joint cartilage stress dynamic monitoring system based on flexible sensing and AI and belongs to the technical field of medical health monitoring.The application solves the problem that the prior art can only measure single-point pressure or strain and cannot obtain full-field three-dimensional stress distribution of a cartilage contact surface, and the system function stops at monitoring and does not form a monitoring-to-warning closed loop, quantifies uncertainty through a Bayesian neural network, provides a reliable basis for clinical decision-making, combines multidimensional biomechanical characteristics and an attention mechanism, accurately focuses on a stress key area, significantly improves the precision and individualization level of joint cartilage health evaluation, helps early detection and prevention of injury, classifies stress abnormalities, implements accurate early warning, and formulates an individualized treatment scheme according to uncertainty results and cartilage health indexes, and customizes a rehabilitation training plan according to user conditions, and improves the scientific nature of joint injury prevention and rehabilitation.

The present application belongs to the field of geological exploration and artificial intelligence technology, and discloses a mineral resource prediction method and system based on a Bayesian physical information neural network. The method comprises the following steps: obtaining and preprocessing multi-source geological data of a study area; constructing a Bayesian neural network based on variational inference; constructing a physical constraint loss with a geological source term; performing three-stage progressive training based on a multi-objective loss function; and performing Monte Carlo sampling prediction and uncertainty quantification. The present application has strong physical interpretability: by embedding a steady-state diffusion equation with a geological source term, the model prediction result conforms to the geological law of ore-forming element migration and enrichment, and the source term clearly corresponds to two geological actions of fracture channel and ore-forming parent rock. The present application realizes the organic combination of data driving and physical driving by using multi-source information such as geochemical data, fracture structure, rock mass distribution and known mine point labels.

A method for training an artificial neural network, particularly a Bayesian neural network, using a training data set includes a step of matching the parameters of the artificial neural network according to a loss function. This loss function includes a first term representing an estimate of the lower bound of the distance between the classification of the training data set by the artificial neural network and the desired classification of the training data set. Furthermore, the loss function includes a second term configured to adjust for differences in random uncertainty in the training data set by different samples passed through the artificial neural network.

The application discloses a kind of based on active learning and man-machine collaborative optimization's massaging robot control method, it is related to artificial intelligence, medical rehabilitation robot and complex contact type operation control cross technical field, this method relies on multimodal perception system, core computing hub and remote man-machine collaborative interface control platform, in turn complete the initial massaging strategy model construction of multi-source heterogeneous expert data, the uncertainty measurement based on bayesian neural network, man-machine collaborative key frame active learning trigger, reinforcement learning reward function design of fusion patient biological feedback and incremental learning of impedance control parameter and force-position hybrid output;The application greatly improves the clinical safety of massaging robot, exponentially reduces data acquisition cost, realizes individual physiological closed-loop rigid-flexible massaging, also overcome the catastrophic forgetting problem of neural network, endow system lifelong learning ability.

This invention relates to the field of textile performance testing and intelligent evaluation technology, specifically providing a home textile sleep-aiding performance evaluation system and construction method based on feature function indicators. The system includes: acquiring five-dimensional feature indicators (tactile, thermal humidity, pressure, interference, and hygiene) through standardized instrument testing, and automatically assigning weights using range normalization and entropy weighting; constructing a physically constrained Bayesian neural network, embedding prior knowledge of materials science and sleep physiology as regularization terms into the loss function, and outputting a sleep-aiding performance score and confidence interval; establishing an adaptive weighted graph convolutional network to achieve knowledge transfer and zero-sample prediction between different home textile products; and fitting the relationship between static indicators and environmental parameters through a dynamic environment adaptive mapping module to output a scenario-based sleep-aiding performance level. This invention integrates instrumental quantitative testing with multi-level neural networks, breaking away from traditional linear regression dependence and achieving rapid, objective, personalized, and scenario-adaptive sleep-aiding performance evaluation.

This invention provides a collision detection method and system for robotic arms based on Bayesian inference, relating to the technical fields of robot safety perception and human-computer interaction. The method includes: constructing a generalized momentum model containing unknown dynamic terms and external joint torques based on a joint spatial dynamics model of the robotic arm; performing probabilistic inference on the unknown dynamic terms using an integrated Bayesian neural network, outputting the predicted mean and variance of the unknown dynamic terms; constructing a data-driven adaptive momentum observer based on the predicted mean and variance to generate an estimated signal of the external joint torques; and performing collision detection based on sparse Bayesian inference on the estimated signal based on the sparsity characteristics of external collision perturbations to generate a collision determination signal. This method solves the technical problems of insufficient sensitivity and high false alarm rate in existing collision detection technologies, achieving improved accuracy in external torque estimation and collision detection sensitivity.

The application discloses an AI-assisted pathological sample diagnosis method and a diagnosis system, relates to the technical field of medical image assistance, and comprises the following steps: acquiring a pathological sample from a digital pathology image library through a data access layer, and acquiring user information corresponding to the pathological sample from a clinical information library; performing multi-scale attention feature extraction and representation learning on the pathological sample through a multi-scale feature extraction network; quantifying the dynamic uncertainty of the pathological sample through a Bayesian neural network; performing multi-modal knowledge fusion; performing diagnosis decision process visualization output through a visualization module; evaluating and verifying the diagnosis system based on the visualization output result; the recognition ability of microstructure in the pathological sample and the confidence evaluation accuracy of the diagnosis result are significantly improved, the explainability and clinical practicability of the system are enhanced, incremental learning and long-term verification are supported, and the robustness, generalization ability and cost-effectiveness of the diagnosis system are effectively improved.

This invention discloses an in-memory computation method for Monte Carlo Bayesian neural networks, belonging to the field of in-memory computing technology. The invention first constructs an in-memory computation architecture for Monte Carlo Bayesian neural networks, using volatile SN nodes to store mask information and controlling the activation of neural network connections corresponding to non-volatile memory, thus achieving the DropConnect operation via hardware. Subsequently, during the repeated erasing and writing of the mask, the high durability, high write / erase speed, and low power consumption of volatile memory are utilized. Parallel simulation computation significantly improves computational energy efficiency and speed.

The application belongs to the technical field of neural networks. Specifically, it is a method for realizing a Bayesian neural network based on random computing by using MRAM. The method is suitable for a Bayesian neural network in a random computing domain. The binary characteristics of MRAM are used to pre-store weight data in MRAM. A triode is used as a switching signal of a circuit to represent data input. PCSA is used as a signal reading mode to obtain a calculation result. The application uses a non-volatile memory device MRAM to design a storage and calculation integrated architecture to realize in-situ storage and calculation of data. Compared with CMOS technology, the application can greatly reduce the calculation power consumption and alleviate the problem of the "memory wall".

The application discloses a structured pruning Bayesian neural network image classification method based on generalized approximate message passing, and comprises the following steps: S1, selecting an image dataset; S2, differentially preprocessing training images of each dataset to generate an image sample set; S3, constructing a BNN network backbone architecture based on Wide ResNet-28x10; S4, constructing a GAMP-BNN model; S5, constructing a GAMP-SPBNN model; S6, removing redundant weights through a weight importance accumulation and gradual pruning rate control strategy; S7, realizing mean and variance estimation of forward transmission and residual information mapping of reverse transmission through a GAMP module, and updating a posterior distribution of network weights in combination with a damping factor; S8, constructing a total loss function comprising a latent loss, a classification loss and a pruning regularization loss; S9, extracting sparse features through a pruning Bayesian convolution layer; and S10, mapping the convolution features to a category space through a pruning Bayesian fully connected layer, and realizing image classification and uncertainty estimation by using a probabilistic output and a multi-model integration strategy.

The application discloses a digital twin of an aircraft, a model construction method, a data processing method and device. The proxy model is deployed in a digital twin system of an electric vertical take-off and landing aircraft. The proxy model construction method comprises: obtaining original training data, each training sample in the original training data comprising flight state parameters, flight environment parameters and structure parameters of the electric vertical take-off and landing aircraft; processing the original training data based on three different fidelity fluid mechanics numerical simulation methods, and correspondingly obtaining low-fidelity training data, medium-fidelity training data and high-fidelity training data, each training sample in each fidelity training data comprising aerodynamic load data simulated by the fluid mechanics numerical simulation method based on the corresponding fidelity according to the training sample. Based on the low-fidelity training data, a Bayesian neural network model is pre-trained to obtain a pre-trained model; based on the medium-fidelity training data and the high-fidelity training data, transfer learning is performed on the pre-trained model to obtain a multi-fidelity proxy model, and the multi-fidelity proxy model predicts flight physical data such as aerodynamic load according to actual flight parameters of the electric vertical take-off and landing aircraft in the digital twin system.

The present application relates to the technical field of electric vehicle power battery health management and intelligent prediction, and discloses a kind of electric vehicle battery health prediction method based on federal migration learning, constructs double branch prediction model, wherein the first branch is inputted with multidimensional feature sequence tensor to carry out multivariate feature coding, the second branch is inputted with original noise SOH sequence based on history, and long-term degradation dependence is extracted based on Mamba and causal attention mechanism;After fusion, the future SOH prediction value and uncertainty are outputted by the Bayesian neural network prediction head;In the training stage, the privacy protection collaborative modeling is carried out between multiple source domain clients using the federal learning framework, and dynamic aggregation is carried out based on the validation loss;For target domain vehicle, unsupervised domain self-adaptive updating is carried out using CORAL feature alignment strategy.In the meantime of protecting data privacy, the accuracy, robustness and migratability of SOH prediction are improved.

The application provides a kind of underwater image recognition method based on bayesian neural network, belong to underwater image recognition technical field;Solve the technical problem that general target recognition method is poor in robustness in visual degradation underwater environment.The method steps include: S1, constructing underwater submarine image dataset;S2, construct underwater image recognition model based on bayesian neural network;S3, use the constructed dataset to train deep learning model, obtain underwater image recognition model;S4, according to the trained prediction model, the image collected when executing underwater task to submersible is predicted and analyzed, and the image target classification result and uncertainty are obtained.

The embodiment of the present application provides a kind of out-of-database target identification method and device, method includes: according to target identification database, construct target identification bayesian neural network, network weight obeys probability distribution;Variational inference method is used to approximate the bayesian network, and reparameterization is expressed;Network is trained until the identification ability meets the demand;When using the trained network to carry out category prediction, input the data to be identified for prediction multiple times, based on the multiple prediction results obtained, calculate target category prediction uncertainty, if uncertainty is greater than the set threshold, then judge target as out-of-database target, if uncertainty is less than the set threshold, then judge target as in-database target, and select the highest average prediction probability category as target prediction category.The embodiment of the present application can effectively automatically identify out-of-database target while ensuring the correct identification of in-database target, and then new target category can be found in time to quickly expand target template library.

This invention discloses a method for evaluating the operational status of a mining drilling rig system based on an echo-state Bayesian neural network. This method addresses the challenges of dynamic feature extraction, model overfitting, and the lack of uncertainty quantification in deterministic assessments of drilling rig status under conditions of strong vibration and high noise in underground environments. It employs an echo-state network with a leakage integral mechanism to suppress noise and extract slowly varying dynamic features from multi-source sensor time-series data, constructing a high-dimensional feature vector through dual-view feature aggregation. A Bayesian neural network with Concrete Dropout is introduced to achieve state probability mapping based on variational inference, and Monte Carlo sampling and prediction entropy are combined to quantify cognitive uncertainty and provide early warning of high-entropy anomalies. This invention effectively filters out high-frequency noise, suppresses overfitting in small samples, and enables proactive early warning of abnormal operating conditions. It achieves high classification accuracy on hard coal seam drilling datasets, and its assessment accuracy and decision reliability are significantly superior to traditional deep learning models.

This invention discloses a capacity optimization configuration method for an integrated photovoltaic-storage-charging microgrid based on a V2G (Vehicle-to-Grid) model, belonging to the field of power system planning and optimization control technology. The method includes constructing a comprehensive operating scenario for a residential photovoltaic-battery storage-electric vehicle system supporting V2G; building a dynamic environment model based on a Bayesian neural network based on the comprehensive operating scenario and its generated historical operating data; constructing a capacity allocation agent based on a deep deterministic policy gradient; generating continuous capacity decisions and evaluating their long-term value; and jointly optimizing the configuration of photovoltaic and energy storage capacities based on the interaction between the dynamic environment model and the capacity allocation agent. This invention significantly reduces the computational complexity of long-term capacity allocation while ensuring accurate characterization of the system's internal power flow and component states. It considers V2G operating modes, electric vehicle availability, and battery degradation factors, improving the practical feasibility of the capacity configuration results.

The invention relates to an electronic circuit for implementing a Bayesian neural network, comprising bit, source, and word lines; andat least one primary branch, each including a primary cell connected between the source and the bit lines and including a primary memory component and a primary switch connected in series,at least one secondary branch, each including a secondary cell connected between the source and the bit lines and including a secondary memory component and a secondary switch connected between them,an accumulation device configured to accumulate a total amount being the sum of a primary amount of charges from a primary cell and a secondary amount of charges from a secondary cell, the primary and secondary amounts being accumulated independently of each other.

This invention discloses a vehicle path tracking control method based on a Bayesian neural network substitution model, belonging to the field of vehicle path tracking control technology. The method includes: constructing a vehicle dynamics model; acquiring pseudo-random front wheel steering angle and accelerator / brake pedal signals, inputting them into the vehicle dynamics model for processing to obtain vehicle motion data; establishing a Bayesian long short-term memory network model based on the vehicle motion data; training a Bayesian neural network substitution model based on the input and output data of the Bayesian long short-term memory network model; constructing a model predictive control framework based on the Bayesian neural network substitution model, and transforming the optimal control problem into a mixed-integer linear programming problem; solving the problem to achieve predictive control of vehicle path tracking. Under complex path conditions, this invention exhibits superior robustness, effectively coping with interference factors such as parameter changes and measurement noise, ensuring the stability and reliability of vehicle path tracking.

This invention provides an intelligent security risk prediction method and system based on a multimodal large model. The method includes the following steps: collecting video, audio, and sensor data, and outputting a unified event token sequence; generating natural language causal descriptions for the event token sequence, constructing an event causal graph with event tokens as nodes, causal descriptions as edges, and outputting logits as edge weights, and dynamically updating the graph structure; obtaining the prior probability of node risk using a Bayesian neural network, fusing the correlation information between nodes through a fusion graph neural network message passing mechanism, and using Do-Calculus intervention loss calibration to obtain the node risk probability; using a Continuous-Time Markov Network to infer future risk trends; outputting a decision report and executing a tiered response; and updating the end-to-end model under privacy protection through federated learning. The beneficial effects of this invention are: achieving cognitive-level risk identification and proactive intervention in complex scenarios; significantly reducing false alarm and false negative rates, and enabling early warning.

The application discloses a cascade hydropower dispatching method and system based on a Bayesian neural network and random optimization, relates to the field of power system optimal dispatching, and specifically relates to: constructing a water inflow scene generation model based on a Bayesian neural network, generating a water inflow scene sequence; considering water flow correlation kinetic energy loss caused by pipe wall friction of a water inflow channel of a hydropower station, constructing a flow-generating capacity loss model, introducing the flow-generating capacity loss model into a day-ahead random optimization dispatching model of a cascade hydropower system, and constructing a random optimization dispatching model of corrected generating capacity; driving the random optimization dispatching model based on the water inflow scene sequence, and adopting a McCormick relaxation method to convert the random optimization dispatching model into a mixed integer linear programming model for solving, and outputting a dispatching optimization scheme of the cascade hydropower system. The application can obtain a robust optimal dispatching decision under uncertain water inflow conditions by combining the probabilistic prediction of the Bayesian neural network with the random optimization dispatching.

This invention discloses a drug sensitivity prediction method and system based on an attention-Bayesian neural network. The method includes: acquiring multi-omics data of a target patient and preprocessing the multi-omics data to obtain initial multi-omics data; processing the initial multi-omics data using an attention weighting mechanism to obtain global fusion features; and processing the global fusion features using a Bayesian neural network to obtain drug sensitivity prediction results and uncertainty scores. Therefore, this invention can adaptively fuse multi-omics data and output prediction results with uncertainty quantification, effectively improving prediction accuracy, robustness, and interpretability, providing a more reliable reference for clinical medication decisions.

The application discloses an aluminum alloy process parameter regulation and control method based on machine learning, relates to the technical field of process parameter regulation and control, and comprises the following steps: after a double-channel data processing path is constructed, a separate full connection layer is used to perform nonlinear physical mapping on a feature subset, a high-order physical state vector is generated, a Bayesian neural network model is trained by taking the high-order physical state vector as input and product performance data as a prediction target, a physical constraint mixed model is obtained, internal gradient information of the physical constraint mixed model is calculated, an uncertainty contribution degree vector is acquired, an optimization target is set, the physical constraint mixed model is taken as an evaluation function, the uncertainty contribution degree vector is taken as a constraint condition, a Pareto optimal solution set is searched for in a process parameter space, and after the Pareto optimal solution set is visualized, optimal process parameter combinations are screened out; and the aluminum alloy hot working process is intelligently, finely and reliably regulated and controlled through the parameter regulation and control method.

This application relates to Bayesian networks in memory. Apparatus and methods may relate to implementing a Bayesian neural network in memory. The Bayesian neural network may be implemented using a resistive memory array. The memory array may include programmable memory cells that can be programmed to store the weights of the Bayesian neural network and perform operations consistent with the Bayesian neural network.

This invention relates to the field of aero-engine maintenance technology, and discloses a multi-objective aero-engine overhaul maintenance decision optimization method and system, including: constructing a decision variable vector containing multi-level variables; constructing a probabilistic quality objective function by calling a physical information Bayesian neural network combined with a risk aversion coefficient; calculating the maintenance cycle and maintenance cost objective functions of the overhaul separately; establishing a constraint set using a reduced-order dynamics model combined with physical boundary thresholds; constructing a mathematical programming model by combining the objective function and the constraint set, and outputting the Pareto front by calling a non-dominated sorting genetic algorithm; calculating the proximity score using a multi-attribute decision method and outputting the solution corresponding to the highest score. This invention quantifies process risks, comprehensively optimizes overhaul quality, cycle, and cost under the premise of ensuring physical safety, and ultimately selects the unique optimal decision solution, improving risk resistance and engineering practicality.

The application provides an ultrasonic ranging interference error prediction method based on an uncertainty perception double-flow model. The spatial position of an ultrasonic emission device and an ultrasonic ranging sequence are extracted as space-time features to construct a double-flow space-time network. The model simultaneously outputs predicted results and two kinds of uncertainties by combining a Bayesian neural network, thereby achieving high robustness and ensuring that the method is more reliable when migrated to different external ultrasonic interference environments. The double-flow neural network simultaneously captures space-time features and fuses predictions by using a long short-term memory network and a double-separated attention graph neural network, thereby comprehensively learning the influence of space-time attributes of the physical world on the prediction accuracy of ultrasonic ranging. The Bayesian network variational inference part introduces a mixed prior, thereby accelerating the convergence speed of the variational distribution and the real distribution in the model training process and realizing faster and more efficient training.

The application provides a highland concrete structure health digital twin system and life prediction method based on multi-element heterogeneous sensing and virtual-real mapping, and belongs to the technical field of concrete structure health monitoring.The application comprises a highland environment structure coupling sensing layer, a 5G edge intelligent transmission layer, a cloud collaborative data processing layer and a physical-data double-driven digital twin layer.The application first introduces an air pressure and resistivity gradient sensor to quantize the multi-field coupling effect of meteorology, freeze-thaw and mechanics;adopts a 5G architecture to realize edge multi-protocol analysis and weak network transmission resistance;finally, based on a physical guided dynamic Bayesian neural network, a deep neural network and a freeze-thaw damage mechanics model are nested, the confidence of the prediction result is quantized through a heterogeneous variance uncertainty output layer, and a virtual-real mapping closed loop of the structure metabolic state is formed, thereby providing a digital twin service covering intelligent monitoring and life prediction for the highland concrete, and breaking through the technical bottleneck of environment-mechanics multi-factor decoupling in traditional monitoring.

The application provides a train diagram elasticity evaluation method based on initial delay influence, and relates to the technical field of train diagram elasticity evaluation, and comprises the following steps: S1, obtaining a target train diagram, wherein target train diagram data comprises train planned operation information; S2, training a Bayesian neural network model for predicting train event delay time based on historical operation data; S3, selecting an initial delay injection event in the target train diagram; S4, simulating the propagation process of each initial delay time in different influence situations in the target train diagram by adopting an autoregressive prediction mode; S5, calculating at least one elasticity evaluation index related to delay propagation influence; and S6, outputting the elasticity evaluation index for evaluating the elasticity of the target train diagram. The application can effectively process delay propagation uncertainty and quantitatively evaluate the elasticity of a train diagram.